Sex ratios and the red dragon: using the Chinese Communist Revolution to explore the effect of the sex ratio on women and children in Taiwan

Abstract

In 1949, Mao Zedong and the People’s Liberation Army defeated the Chinese Nationalist Army. Hundreds of thousands of mainland Chinese fled to the island of Taiwan. In this paper, I use the demographic consequences of the Chinese Communist Revolution and subsequent Taiwanese military policy to identify the effect of the marriage market sex ratio on women and children in Taiwan. I find that as the sex ratio rises, the bride price relative to the dowry increases, the fraction of female children in a family increases, the total number of children in a family decreases, and human capital investments in children increase.

Population data appendix

In this data appendix, I describe the procedure that I follow to estimate the marriage market sex ratio and other population statistics. I use population data collected through the household registration system. In Taiwan, every birth, death, marriage, and intercounty change of residence must be registered with the government within a couple of weeks. The household registration system is comprehensive. Statistics based on the household registration system compare well with Taiwanese census figures (Republic of China 1961). Annual Taiwan-level statistics on population and death rates by gender and age for 1950–1989 come from the Statistical Abstract (Republic of China 1974a) and Statistical Yearbook (Republic of China 1975a, 1999, 2004). Annual county-level statistics on population by gender and age for 1961–1989 come from the Demographic Fact Book (Republic of China 1961, 1963–1968, 1971–1973, 1974b, 1975b, 1976–1989); 1961 is the earliest year for which the county-level data are available.

For the years 1950–1968, the raw Taiwan-level data exclude all men enlisted in the military. For the years 1961–1968, the raw county-level data exclude mainlanders enlisted in the military. All soldiers are counted starting in 1969. Prior to 1969, soldiers are counted only when they retire. Lin (2002) speculates that the Taiwanese government does not report the number of soldiers during this period to keep its military capabilities secret from China. I calculate marriage market sex ratios that exclude mainlanders in the military. For this reason, I must distinguish between civilians and soldiers, as well as between mainlanders and native Taiwanese in the military. Specifically, I add to the raw Taiwanese population data the estimated number of native Taiwanese in the military prior to 1969; subtract from the raw Taiwanese population data the estimated number of mainlanders in the military after 1969; and subtract from the raw county population data the estimated number of mainlanders in the military after 1969. My estimates compare well with other studies (Bullard 1997; Lin 2002; Roy 2003; Wang 1993b). See the discussion in Section 2.

The estimation procedure consists in four steps. First, I estimate the total civilian and military population of Taiwan by gender and age for 1950–1968. Going backwards in time, I use gender-specific and age-specific death rates and post-1969 population statistics, which include the military, to estimate the total number of men and women living in Taiwan by age in each year prior to 1969. I assume that there is no immigration to or emigration from the island and that civilian and military death rates are identical. As I have discussed previously, there is virtually no immigration or emigration during the period. Overall death rates are quite low for young and middle-aged adults. Moreover, the military is not involved in any significant combat. Hence, the population estimates are not sensitive to the assumptions about immigration and death rates. I then subtract the raw Taiwan-level population statistics from my estimates of the total population to obtain estimates of the military population by gender and age.

Second, I distinguish between mainlanders and native Taiwanese in the military for 1950–1968. Native Taiwanese males aged 20–24 began to enlist in the military in 1956 when the government established the compulsory military service system (Bullard 1997; Hsu 2003). By this time, mainlanders in the military were older than 20–24 years old. I follow the cohorts of native Taiwanese and mainlanders enlisted in the military over time. New recruits are native Taiwanese, and older soldiers are mainlanders. With these estimates, I add to the raw Taiwanese population data the number of native Taiwanese in the military prior to 1969.

Third, I estimate the number of mainlanders in the military post-1969. I calculate the age-specific rate of retirement for mainlanders in the military prior to 1969. Using these rates of retirement and estimates of the number of mainlanders in the military prior to 1969, I estimate the number of mainlanders in the military post-1969. The number of mainlanders in the military after 1969 is small, so these estimates are not sensitive to the assumptions. I subtract from the raw Taiwanese population data the number of mainlanders in the military after 1969.

Fourth, I estimate the number of mainlanders in the military post-1969 by county. For each county–age cell, I calculate how many men one would have to add to the 1968 population to make the 1968 county–age sex ratio equal to the corresponding 1969 sex ratio, which already includes the military. With this, I calculate the age-specific distribution of mainlanders in the military across countries in 1968. Assuming this distribution is fixed across time, I multiply the total number of mainlanders in the military by each county’s proportion for each age and year. I subtract from the raw county population data the number of mainlanders in the military after 1969. These estimates are not sensitive to the assumptions because the total number of mainlanders in the military post-1969 is small.